Overall interaction method of joint learning platform

Abstract

The invention discloses an overall interaction method of a joint learning platform, which belongs to the technical field of joint learning platforms and comprises the steps of S1, resource registration; s2, querying an ecological library: querying model library related information from the basic AI-ecological model library by a user through an industry front-end background, and querying industry data from an industry AI-database; s3, creating a task; s4, querying a task; s5, adding a task; s6, the Agent is started; and S7, carrying out joint training. The operations of resource registration, ecological library query, task creation, task query, task addition and Agent starting can be performed through an industry front-end background, and each process of the joint learning platform establishes interactive connection with a user, so that comprehensive interactive learning is realized on the whole; and the joint training model is continuously improved, the joint learning precision is improved, and the joint learning efficiency and precision are further improved.

Description

technical field

[0001] The invention relates to the technical field of joint learning platforms, in particular to an overall interaction method of joint learning platforms. Background technique

[0002] Bayesian logic combines generative models and probability theory to calculate the true likelihood of a particular answer given data, bottom-up deep learning starts with data, not theory, and involves labeling large amounts of data ("correct" ” and “error” data) to identify associations and lay the groundwork for pattern recognition. It can even learn unsupervised, detect patterns in unlabeled data, and identify clusters (factor analysis). The era of renewed interest in machine learning technology was followed by the era of deep learning technology. Recent years have led us to the next in-depth study of intelligent and / or neuromorphic computing, namely federated learning techniques.

[0003] With machine learning, a human can enter the input and the desired output. After t...

Examples

Embodiment 1

[0069] see figure 1 , an overall interaction method for a federated learning platform, including the following steps:

[0070] S1: Resource registration, the user uploads local resources to the computing decision-making module, and receives registration feedback from the computing and decision-making module;

[0071] S2: Query the ecological database, the user queries the basic AI-ecological model database for relevant information of the model database through the industry front-end and back-end, and queries the industry data from the industry AI-database;

[0072] S3: Create a task, perform initial task configuration to the industry training configuration through the industry front-end and background, and submit the task;

[0073] S4: Query tasks, query joint training tasks from the basic AI-joint learning engine through the industry front-end and back-end;

[0074] S5: Add tasks, and add joint learning tasks to the basic AI-joint learning engine through the industry front-...

Embodiment 2

[0102] see Figure 8 , the joint training includes the following steps:

[0103] S701: Requesting data, the Agent obtains data directly from the library or locally according to the metadata information provided by the industry user, and integrates it into a training data set;

[0104]S702: Obtain the monitoring information, send a request to the platform resource controller to obtain the monitoring information, and the Agent judges whether to perform this joint iteration process according to the monitoring information;

[0105] S703: This round of iterative training, when the monitored information meets the training standard, the Agent performs this iterative training, uploads the model trained in this round to fl-engine, and downloads the aggregated model from fl-engine;

[0106] S704: The interaction of monitoring information, the Agent feeds back the monitoring information obtained from the platform resource controller before the iteration, the heartbeat information obtain...

Embodiment 3

[0109] see Figure 9 , the joint training includes the following steps:

[0110] S701: Request data. After the Agent is started, it first calls the industry backend API interface to obtain a certain training data set according to the parameters. After receiving the request, the industry backend queries the corresponding database or local server to find the corresponding training data, and stores the training data. Return to Agent;

[0111] S702: Obtain the monitoring information, send a request to the platform resource controller to obtain the monitoring information, and the Agent judges whether to perform this joint iteration process according to the monitoring information;

[0112] S703: This round of iterative training, when the monitored information meets the training standard, the Agent performs this iterative training, uploads the model trained in this round to fl-engine, and downloads the aggregated model from fl-engine;

[0113] S704: The interaction of monitoring in...

technical field

[0001] The invention relates to the technical field of joint learning platforms, in particular to an overall interaction method of joint learning platforms. Background technique

[0002] Bayesian logic combines generative models and probability theory to calculate the true likelihood of a particular answer given data, bottom-up deep learning starts with data, not theory, and involves labeling large amounts of data ("correct" ” and “error” data) to identify associations and lay the groundwork for pattern recognition. It can even learn unsupervised, detect patterns in unlabeled data, and identify clusters (factor analysis). The era of renewed interest in machine learning technology was followed by the era of deep learning technology. Recent years have led us to the next in-depth study of intelligent and / or neuromorphic computing, namely federated learning techniques.

[0003] With machine learning, a human can enter the input and the desired output. After t...

Examples

Embodiment 1

[0069] see figure 1 , an overall interaction method for a federated learning platform, including the following steps:

[0070] S1: Resource registration, the user uploads local resources to the computing decision-making module, and receives registration feedback from the computing and decision-making module;

[0071] S2: Query the ecological database, the user queries the basic AI-ecological model database for relevant information of the model database through the industry front-end and back-end, and queries the industry data from the industry AI-database;

[0072] S3: Create a task, perform initial task configuration to the industry training configuration through the industry front-end and background, and submit the task;

[0073] S4: Query tasks, query joint training tasks from the basic AI-joint learning engine through the industry front-end and back-end;

[0074] S5: Add tasks, and add joint learning tasks to the basic AI-joint learning engine through the industry front-...

Embodiment 2

[0102] see Figure 8 , the joint training includes the following steps:

[0103] S701: Requesting data, the Agent obtains data directly from the library or locally according to the metadata information provided by the industry user, and integrates it into a training data set;

[0104]S702: Obtain the monitoring information, send a request to the platform resource controller to obtain the monitoring information, and the Agent judges whether to perform this joint iteration process according to the monitoring information;

[0105] S703: This round of iterative training, when the monitored information meets the training standard, the Agent performs this iterative training, uploads the model trained in this round to fl-engine, and downloads the aggregated model from fl-engine;

[0106] S704: The interaction of monitoring information, the Agent feeds back the monitoring information obtained from the platform resource controller before the iteration, the heartbeat information obtain...

Embodiment 3

[0109] see Figure 9 , the joint training includes the following steps:

[0110] S701: Request data. After the Agent is started, it first calls the industry backend API interface to obtain a certain training data set according to the parameters. After receiving the request, the industry backend queries the corresponding database or local server to find the corresponding training data, and stores the training data. Return to Agent;

[0111] S702: Obtain the monitoring information, send a request to the platform resource controller to obtain the monitoring information, and the Agent judges whether to perform this joint iteration process according to the monitoring information;

[0112] S703: This round of iterative training, when the monitored information meets the training standard, the Agent performs this iterative training, uploads the model trained in this round to fl-engine, and downloads the aggregated model from fl-engine;

[0113] S704: The interaction of monitoring in...

Images